AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: The Harvard Study Looks At AI's Potential To Ease App Switching on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get audio and creator gear delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

A Harvard Business School preprint analysing 103 million application events from 1,017 workers in India found an average of 908 app switches per workday. Workers switched less in the ten app uses after an AI use, but made about 200 more switches on days they used AI; the observational study cannot establish why.

A Harvard Business School study of more than 103 million application events found that workers switched apps less in the sequence immediately after using generative AI, yet made about 200 more switches on AI-use days overall. The findings describe associations, not cause and effect, and come from a preprint that remains under peer review.

Researchers Sumer S. Vaid and Ashley V. Whillans analysed computer activity from 1,017 office workers at eight organisations in India. The records, gathered from November 2025 to January 2026 through workforce-analytics software, covered 103 million application events and 62,693 person-days. The researchers counted a switch when a worker moved from one application to another; opening a different document in the same app did not count as a new application use.

Across the dataset, workers averaged 908.3 app switches per day, while the median was 814. The middle half of employees recorded between 547 and 1,130 switches per workday. The study also measured time spent in an application before moving to another: the mean was 27.8 seconds, while the median was 3.3 seconds. Those figures describe recorded app activity, not necessarily sustained attention or the quality of someone’s work.

The AI results differed depending on the time window. In the ten application uses after an AI use, workers used fewer apps, switched less, stayed about half a second longer in each app and moved through apps more predictably. But on days when they used AI, workers made roughly 200 more switches per day; more minutes of AI use were also associated with more fragmented activity. The paper names five AI tools in the data, including Anthropic’s Claude.

At a glance
reportWhen: Preprint posted July 7, 2026, revised A…
The developmentA Harvard Business School preprint reports that AI use is associated with fewer app switches immediately afterward but more switches over the course of AI-use days.

App switching and AI · one moment

What happens around one AI query?

The study zoomed in on every AI use: the ten app uses before it, and the ten after. Up close, work calms down a little. Zoom out to the whole day, and the picture changes.

10 app uses before AI median 7 seconds 10 app uses after Whole AI day: about 200 more switches
  1. 1 · A workday of switchingWorkers changed apps 908 times a day on average. One app held their focus for 27.8 seconds on average, 3.3 seconds at the median.
  2. 2 · AI comes in at a busy momentPeople used AI a median of 13 times a day, for a median of 7 seconds. Fragmentation peaked in the hour of AI use.
  3. 3 · The next ten app usesFewer different apps, fewer switches, about half a second longer per app, more predictable. Real, but small.
  4. 4 · Zoom out: the whole dayOn days with AI use, the same people made about 200 more switches. Whether AI calms work or busy moments bring AI: unknown.

Source: Sumer S. Vaid, Ashley V. Whillans (Harvard Business School), “Digital Fragmentation and Generative AI Use Across 103 Million Application Events”, arXiv:2607.06681v2, preprint under review. Blocks are schematic, not to scale.

Graphic: Powered by Thorsten Meyer AI

AI and the Shape of a Workday

The findings challenge a simple pitch that adding AI will automatically reduce workplace interruptions. They suggest that the relationship may look different across a short sequence of tasks and across an entire day: fewer switches immediately after AI use can coexist with more switching on days when AI is used.

For workers and employers, that distinction matters when evaluating claims that AI tools improve focus. The paper does not measure whether the tools helped people finish work faster, produce better results or feel less interrupted. Its data therefore cannot show that AI improved productivity or that greater switching harmed it. It offers a measure of recorded app movement, not a verdict on the value of AI at work.

The study also reports that day-to-day variation within the same person accounted for 44.6% of variation in fragmentation, compared with 35.8% associated with differences between people and 19.6% between organisations. That pattern suggests that work conditions can vary substantially from one day to another, but it does not identify what caused those changes. Employers should not treat the results as proof that AI is either a focus aid or a distraction.

Products Worth Considering

As an affiliate, we earn on qualifying purchases.

How the Study Tracked App Use

The paper, titled Digital Fragmentation and Generative AI Use Across 103 Million Application Events, was written by Vaid of Harvard Business School’s Artificial Intelligence Institute and Whillans of its Negotiation, Organizations and Markets Unit. It was first posted on arXiv on July 7, 2026, and revised on August 7. The manuscript identifies itself as a preprint and says its findings may change through peer review.

The sample covered eight Indian organisations in software, accounting, financial services, e-publishing, law and marketing. A commercial service called Flowace supplied the activity records. Its software recorded when the active application changed, along with information such as the app name and the start and end time. Employers, rather than the researchers, determined which applications counted as productive.

The study says Harvard’s human-subjects committee approved the protocol and that researchers received anonymised data under a data-use agreement. The authors reported no external funding and said Flowace had no role in study design, analysis or interpretation. One author disclosed equity in an unrelated private company that uses AI to create enterprise digital twins. The sample represents monitored computer activity at these eight organisations; it is not a survey of all workers or devices.

“The manuscript is currently under review, and the findings may change based on the peer review process.”

— The paper’s authors

What the App Records Cannot Show

The data do not establish whether AI use caused workers to switch apps more often, whether people used AI because their day was already fragmented or consolidating, or whether both processes occurred. The authors explicitly leave those possibilities open. The analysis is observational, not an experiment that assigned workers to use or avoid AI.

It is also unclear whether the measured switching affected work quality, output, stress or productivity; those outcomes were not established by the app-event counts described in the paper. The sample comes from eight organisations in India and does not cover phone use, other countries or every kind of workplace. The authors say the results may change through peer review, and comparisons with earlier studies of time spent on tasks should be treated cautiously because the methods differ.

Peer Review and Further Evidence

The paper’s next major milestone is peer review. Until that process is complete, the reported results should be read as preliminary, particularly because the link between AI use and daily switching has no established direction of cause.

Further research would need to test whether AI itself changes switching, and whether any change affects work outcomes. That could include studies that compare similar workdays with and without AI use while measuring task completion and quality, alongside app activity. For now, the study supports a limited conclusion: recorded switching was lower in the short sequence after AI use but higher on AI-use days overall in this particular sample.

Key Questions

Did the study find that AI reduces app switching?

Not overall. Workers switched less in the ten app uses after an AI use, but made about 200 more switches on days when they used AI. The results vary by time window.

Does the study prove that AI causes more switching?

No. The study is observational and cannot establish whether AI use caused the change, whether workers used AI on already fragmented days, or whether both patterns contributed.

Who and what did the researchers study?

The researchers analysed computer-application records from 1,017 employees at eight organisations in India, collected from November 2025 through January 2026 using Flowace workforce-analytics software.

What does the figure of 908 switches mean?

It is the study’s mean number of changes between different applications per worker per day. The median was 814. A change between two documents in the same app did not count as a switch.

Has the Harvard paper been peer reviewed?

No. The paper was posted as a preprint and says it is under review; its authors warn that findings may change during peer review.

Source: ThorstenMeyerAI.com

You May Also Like

How AI Is Testing Britain’s Bureaucracy With More Paperwork

Sky News reports rising written complaints and requests since ChatGPT launched, as officials warn AI is adding to workloads and many claims may be valid.

How AI Transcription Software Can Help With Meetings And Field Work

A product roundup compares 11 transcription tools for meetings, dictation and field recordings, while flagging workflow, privacy and accuracy questions.

Some Workers Lose Jobs To AI, Then Get A Call To Return

Surveys report workers and HR leaders returning to roles cut for AI, but the figures are self-reported and the clearest documented reversal involved 45 jobs.

Claude Meetings Is An AI Participant For Google Meet. Are You Ready?

Anthropic’s beta adds Claude to Google Meet for live notes, answers and documents, with different privacy defaults for Team and Enterprise plans.